Preprocessor.h revision f4f6c9db68465b886ec2e596feaa6ecc782395a4
1//===--- Preprocessor.h - C Language Family Preprocessor --------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file is distributed under the University of Illinois Open Source 6// License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the Preprocessor interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_LEX_PREPROCESSOR_H 15#define LLVM_CLANG_LEX_PREPROCESSOR_H 16 17#include "clang/Lex/Lexer.h" 18#include "clang/Lex/PTHLexer.h" 19#include "clang/Lex/PPCallbacks.h" 20#include "clang/Lex/TokenLexer.h" 21#include "clang/Lex/PTHManager.h" 22#include "clang/Basic/Builtins.h" 23#include "clang/Basic/Diagnostic.h" 24#include "clang/Basic/IdentifierTable.h" 25#include "clang/Basic/SourceLocation.h" 26#include "llvm/ADT/DenseMap.h" 27#include "llvm/ADT/OwningPtr.h" 28#include "llvm/ADT/SmallVector.h" 29#include "llvm/Support/Allocator.h" 30#include <vector> 31 32namespace clang { 33 34class SourceManager; 35class ExternalPreprocessorSource; 36class FileManager; 37class FileEntry; 38class HeaderSearch; 39class PragmaNamespace; 40class PragmaHandler; 41class CommentHandler; 42class ScratchBuffer; 43class TargetInfo; 44class PPCallbacks; 45class DirectoryLookup; 46class PreprocessingRecord; 47 48/// Preprocessor - This object engages in a tight little dance with the lexer to 49/// efficiently preprocess tokens. Lexers know only about tokens within a 50/// single source file, and don't know anything about preprocessor-level issues 51/// like the #include stack, token expansion, etc. 52/// 53class Preprocessor { 54 Diagnostic *Diags; 55 LangOptions Features; 56 const TargetInfo &Target; 57 FileManager &FileMgr; 58 SourceManager &SourceMgr; 59 ScratchBuffer *ScratchBuf; 60 HeaderSearch &HeaderInfo; 61 62 /// \brief External source of macros. 63 ExternalPreprocessorSource *ExternalSource; 64 65 /// PTH - An optional PTHManager object used for getting tokens from 66 /// a token cache rather than lexing the original source file. 67 llvm::OwningPtr<PTHManager> PTH; 68 69 /// BP - A BumpPtrAllocator object used to quickly allocate and release 70 /// objects internal to the Preprocessor. 71 llvm::BumpPtrAllocator BP; 72 73 /// Identifiers for builtin macros and other builtins. 74 IdentifierInfo *Ident__LINE__, *Ident__FILE__; // __LINE__, __FILE__ 75 IdentifierInfo *Ident__DATE__, *Ident__TIME__; // __DATE__, __TIME__ 76 IdentifierInfo *Ident__INCLUDE_LEVEL__; // __INCLUDE_LEVEL__ 77 IdentifierInfo *Ident__BASE_FILE__; // __BASE_FILE__ 78 IdentifierInfo *Ident__TIMESTAMP__; // __TIMESTAMP__ 79 IdentifierInfo *Ident__COUNTER__; // __COUNTER__ 80 IdentifierInfo *Ident_Pragma, *Ident__VA_ARGS__; // _Pragma, __VA_ARGS__ 81 IdentifierInfo *Ident__has_feature; // __has_feature 82 IdentifierInfo *Ident__has_builtin; // __has_builtin 83 IdentifierInfo *Ident__has_include; // __has_include 84 IdentifierInfo *Ident__has_include_next; // __has_include_next 85 86 SourceLocation DATELoc, TIMELoc; 87 unsigned CounterValue; // Next __COUNTER__ value. 88 89 enum { 90 /// MaxIncludeStackDepth - Maximum depth of #includes. 91 MaxAllowedIncludeStackDepth = 200 92 }; 93 94 // State that is set before the preprocessor begins. 95 bool KeepComments : 1; 96 bool KeepMacroComments : 1; 97 98 // State that changes while the preprocessor runs: 99 bool InMacroArgs : 1; // True if parsing fn macro invocation args. 100 101 /// Whether the preprocessor owns the header search object. 102 bool OwnsHeaderSearch : 1; 103 104 /// DisableMacroExpansion - True if macro expansion is disabled. 105 bool DisableMacroExpansion : 1; 106 107 /// \brief Whether we have already loaded macros from the external source. 108 mutable bool ReadMacrosFromExternalSource : 1; 109 110 /// Identifiers - This is mapping/lookup information for all identifiers in 111 /// the program, including program keywords. 112 mutable IdentifierTable Identifiers; 113 114 /// Selectors - This table contains all the selectors in the program. Unlike 115 /// IdentifierTable above, this table *isn't* populated by the preprocessor. 116 /// It is declared/instantiated here because it's role/lifetime is 117 /// conceptually similar the IdentifierTable. In addition, the current control 118 /// flow (in clang::ParseAST()), make it convenient to put here. 119 /// FIXME: Make sure the lifetime of Identifiers/Selectors *isn't* tied to 120 /// the lifetime fo the preprocessor. 121 SelectorTable Selectors; 122 123 /// BuiltinInfo - Information about builtins. 124 Builtin::Context BuiltinInfo; 125 126 /// PragmaHandlers - This tracks all of the pragmas that the client registered 127 /// with this preprocessor. 128 PragmaNamespace *PragmaHandlers; 129 130 /// \brief Tracks all of the comment handlers that the client registered 131 /// with this preprocessor. 132 std::vector<CommentHandler *> CommentHandlers; 133 134 /// \brief The file that we're performing code-completion for, if any. 135 const FileEntry *CodeCompletionFile; 136 137 /// \brief The number of bytes that we will initially skip when entering the 138 /// main file, which is used when loading a precompiled preamble, along 139 /// with a flag that indicates whether skipping this number of bytes will 140 /// place the lexer at the start of a line. 141 std::pair<unsigned, bool> SkipMainFilePreamble; 142 143 /// CurLexer - This is the current top of the stack that we're lexing from if 144 /// not expanding a macro and we are lexing directly from source code. 145 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 146 llvm::OwningPtr<Lexer> CurLexer; 147 148 /// CurPTHLexer - This is the current top of stack that we're lexing from if 149 /// not expanding from a macro and we are lexing from a PTH cache. 150 /// Only one of CurLexer, CurPTHLexer, or CurTokenLexer will be non-null. 151 llvm::OwningPtr<PTHLexer> CurPTHLexer; 152 153 /// CurPPLexer - This is the current top of the stack what we're lexing from 154 /// if not expanding a macro. This is an alias for either CurLexer or 155 /// CurPTHLexer. 156 PreprocessorLexer *CurPPLexer; 157 158 /// CurLookup - The DirectoryLookup structure used to find the current 159 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 160 /// implement #include_next and find directory-specific properties. 161 const DirectoryLookup *CurDirLookup; 162 163 /// CurTokenLexer - This is the current macro we are expanding, if we are 164 /// expanding a macro. One of CurLexer and CurTokenLexer must be null. 165 llvm::OwningPtr<TokenLexer> CurTokenLexer; 166 167 /// IncludeMacroStack - This keeps track of the stack of files currently 168 /// #included, and macros currently being expanded from, not counting 169 /// CurLexer/CurTokenLexer. 170 struct IncludeStackInfo { 171 Lexer *TheLexer; 172 PTHLexer *ThePTHLexer; 173 PreprocessorLexer *ThePPLexer; 174 TokenLexer *TheTokenLexer; 175 const DirectoryLookup *TheDirLookup; 176 177 IncludeStackInfo(Lexer *L, PTHLexer* P, PreprocessorLexer* PPL, 178 TokenLexer* TL, const DirectoryLookup *D) 179 : TheLexer(L), ThePTHLexer(P), ThePPLexer(PPL), TheTokenLexer(TL), 180 TheDirLookup(D) {} 181 }; 182 std::vector<IncludeStackInfo> IncludeMacroStack; 183 184 /// Callbacks - These are actions invoked when some preprocessor activity is 185 /// encountered (e.g. a file is #included, etc). 186 PPCallbacks *Callbacks; 187 188 /// Macros - For each IdentifierInfo with 'HasMacro' set, we keep a mapping 189 /// to the actual definition of the macro. 190 llvm::DenseMap<IdentifierInfo*, MacroInfo*> Macros; 191 192 /// MICache - A "freelist" of MacroInfo objects that can be reused for quick 193 /// allocation. 194 /// FIXME: why not use a singly linked list? 195 std::vector<MacroInfo*> MICache; 196 197 /// MacroArgCache - This is a "freelist" of MacroArg objects that can be 198 /// reused for quick allocation. 199 MacroArgs *MacroArgCache; 200 friend class MacroArgs; 201 202 // Various statistics we track for performance analysis. 203 unsigned NumDirectives, NumIncluded, NumDefined, NumUndefined, NumPragma; 204 unsigned NumIf, NumElse, NumEndif; 205 unsigned NumEnteredSourceFiles, MaxIncludeStackDepth; 206 unsigned NumMacroExpanded, NumFnMacroExpanded, NumBuiltinMacroExpanded; 207 unsigned NumFastMacroExpanded, NumTokenPaste, NumFastTokenPaste; 208 unsigned NumSkipped; 209 210 /// Predefines - This string is the predefined macros that preprocessor 211 /// should use from the command line etc. 212 std::string Predefines; 213 214 /// TokenLexerCache - Cache macro expanders to reduce malloc traffic. 215 enum { TokenLexerCacheSize = 8 }; 216 unsigned NumCachedTokenLexers; 217 TokenLexer *TokenLexerCache[TokenLexerCacheSize]; 218 219 /// \brief A record of the macro definitions and instantiations that 220 /// occurred during preprocessing. 221 /// 222 /// This is an optional side structure that can be enabled with 223 /// \c createPreprocessingRecord() prior to preprocessing. 224 PreprocessingRecord *Record; 225 226private: // Cached tokens state. 227 typedef llvm::SmallVector<Token, 1> CachedTokensTy; 228 229 /// CachedTokens - Cached tokens are stored here when we do backtracking or 230 /// lookahead. They are "lexed" by the CachingLex() method. 231 CachedTokensTy CachedTokens; 232 233 /// CachedLexPos - The position of the cached token that CachingLex() should 234 /// "lex" next. If it points beyond the CachedTokens vector, it means that 235 /// a normal Lex() should be invoked. 236 CachedTokensTy::size_type CachedLexPos; 237 238 /// BacktrackPositions - Stack of backtrack positions, allowing nested 239 /// backtracks. The EnableBacktrackAtThisPos() method pushes a position to 240 /// indicate where CachedLexPos should be set when the BackTrack() method is 241 /// invoked (at which point the last position is popped). 242 std::vector<CachedTokensTy::size_type> BacktrackPositions; 243 244public: 245 Preprocessor(Diagnostic &diags, const LangOptions &opts, 246 const TargetInfo &target, 247 SourceManager &SM, HeaderSearch &Headers, 248 IdentifierInfoLookup *IILookup = 0, 249 bool OwnsHeaderSearch = false); 250 251 ~Preprocessor(); 252 253 Diagnostic &getDiagnostics() const { return *Diags; } 254 void setDiagnostics(Diagnostic &D) { Diags = &D; } 255 256 const LangOptions &getLangOptions() const { return Features; } 257 const TargetInfo &getTargetInfo() const { return Target; } 258 FileManager &getFileManager() const { return FileMgr; } 259 SourceManager &getSourceManager() const { return SourceMgr; } 260 HeaderSearch &getHeaderSearchInfo() const { return HeaderInfo; } 261 262 IdentifierTable &getIdentifierTable() { return Identifiers; } 263 SelectorTable &getSelectorTable() { return Selectors; } 264 Builtin::Context &getBuiltinInfo() { return BuiltinInfo; } 265 llvm::BumpPtrAllocator &getPreprocessorAllocator() { return BP; } 266 267 void setPTHManager(PTHManager* pm); 268 269 PTHManager *getPTHManager() { return PTH.get(); } 270 271 void setExternalSource(ExternalPreprocessorSource *Source) { 272 ExternalSource = Source; 273 } 274 275 ExternalPreprocessorSource *getExternalSource() const { 276 return ExternalSource; 277 } 278 279 /// SetCommentRetentionState - Control whether or not the preprocessor retains 280 /// comments in output. 281 void SetCommentRetentionState(bool KeepComments, bool KeepMacroComments) { 282 this->KeepComments = KeepComments | KeepMacroComments; 283 this->KeepMacroComments = KeepMacroComments; 284 } 285 286 bool getCommentRetentionState() const { return KeepComments; } 287 288 /// isCurrentLexer - Return true if we are lexing directly from the specified 289 /// lexer. 290 bool isCurrentLexer(const PreprocessorLexer *L) const { 291 return CurPPLexer == L; 292 } 293 294 /// getCurrentLexer - Return the current lexer being lexed from. Note 295 /// that this ignores any potentially active macro expansions and _Pragma 296 /// expansions going on at the time. 297 PreprocessorLexer *getCurrentLexer() const { return CurPPLexer; } 298 299 /// getCurrentFileLexer - Return the current file lexer being lexed from. 300 /// Note that this ignores any potentially active macro expansions and _Pragma 301 /// expansions going on at the time. 302 PreprocessorLexer *getCurrentFileLexer() const; 303 304 /// getPPCallbacks/addPPCallbacks - Accessors for preprocessor callbacks. 305 /// Note that this class takes ownership of any PPCallbacks object given to 306 /// it. 307 PPCallbacks *getPPCallbacks() const { return Callbacks; } 308 void addPPCallbacks(PPCallbacks *C) { 309 if (Callbacks) 310 C = new PPChainedCallbacks(C, Callbacks); 311 Callbacks = C; 312 } 313 314 /// getMacroInfo - Given an identifier, return the MacroInfo it is #defined to 315 /// or null if it isn't #define'd. 316 MacroInfo *getMacroInfo(IdentifierInfo *II) const { 317 return II->hasMacroDefinition() ? Macros.find(II)->second : 0; 318 } 319 320 /// setMacroInfo - Specify a macro for this identifier. 321 /// 322 void setMacroInfo(IdentifierInfo *II, MacroInfo *MI); 323 324 /// macro_iterator/macro_begin/macro_end - This allows you to walk the current 325 /// state of the macro table. This visits every currently-defined macro. 326 typedef llvm::DenseMap<IdentifierInfo*, 327 MacroInfo*>::const_iterator macro_iterator; 328 macro_iterator macro_begin(bool IncludeExternalMacros = true) const; 329 macro_iterator macro_end(bool IncludeExternalMacros = true) const; 330 331 const std::string &getPredefines() const { return Predefines; } 332 /// setPredefines - Set the predefines for this Preprocessor. These 333 /// predefines are automatically injected when parsing the main file. 334 void setPredefines(const char *P) { Predefines = P; } 335 void setPredefines(const std::string &P) { Predefines = P; } 336 337 /// getIdentifierInfo - Return information about the specified preprocessor 338 /// identifier token. The version of this method that takes two character 339 /// pointers is preferred unless the identifier is already available as a 340 /// string (this avoids allocation and copying of memory to construct an 341 /// std::string). 342 IdentifierInfo *getIdentifierInfo(llvm::StringRef Name) const { 343 return &Identifiers.get(Name); 344 } 345 346 /// AddPragmaHandler - Add the specified pragma handler to the preprocessor. 347 /// If 'Namespace' is non-null, then it is a token required to exist on the 348 /// pragma line before the pragma string starts, e.g. "STDC" or "GCC". 349 void AddPragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 350 void AddPragmaHandler(PragmaHandler *Handler) { 351 AddPragmaHandler(llvm::StringRef(), Handler); 352 } 353 354 /// RemovePragmaHandler - Remove the specific pragma handler from 355 /// the preprocessor. If \arg Namespace is non-null, then it should 356 /// be the namespace that \arg Handler was added to. It is an error 357 /// to remove a handler that has not been registered. 358 void RemovePragmaHandler(llvm::StringRef Namespace, PragmaHandler *Handler); 359 void RemovePragmaHandler(PragmaHandler *Handler) { 360 RemovePragmaHandler(llvm::StringRef(), Handler); 361 } 362 363 /// \brief Add the specified comment handler to the preprocessor. 364 void AddCommentHandler(CommentHandler *Handler); 365 366 /// \brief Remove the specified comment handler. 367 /// 368 /// It is an error to remove a handler that has not been registered. 369 void RemoveCommentHandler(CommentHandler *Handler); 370 371 /// \brief Retrieve the preprocessing record, or NULL if there is no 372 /// preprocessing record. 373 PreprocessingRecord *getPreprocessingRecord() const { return Record; } 374 375 /// \brief Create a new preprocessing record, which will keep track of 376 /// all macro expansions, macro definitions, etc. 377 void createPreprocessingRecord(); 378 379 /// EnterMainSourceFile - Enter the specified FileID as the main source file, 380 /// which implicitly adds the builtin defines etc. 381 void EnterMainSourceFile(); 382 383 /// EndSourceFile - Inform the preprocessor callbacks that processing is 384 /// complete. 385 void EndSourceFile(); 386 387 /// EnterSourceFile - Add a source file to the top of the include stack and 388 /// start lexing tokens from it instead of the current buffer. Emit an error 389 /// and don't enter the file on error. 390 void EnterSourceFile(FileID CurFileID, const DirectoryLookup *Dir, 391 SourceLocation Loc); 392 393 /// EnterMacro - Add a Macro to the top of the include stack and start lexing 394 /// tokens from it instead of the current buffer. Args specifies the 395 /// tokens input to a function-like macro. 396 /// 397 /// ILEnd specifies the location of the ')' for a function-like macro or the 398 /// identifier for an object-like macro. 399 void EnterMacro(Token &Identifier, SourceLocation ILEnd, MacroArgs *Args); 400 401 /// EnterTokenStream - Add a "macro" context to the top of the include stack, 402 /// which will cause the lexer to start returning the specified tokens. 403 /// 404 /// If DisableMacroExpansion is true, tokens lexed from the token stream will 405 /// not be subject to further macro expansion. Otherwise, these tokens will 406 /// be re-macro-expanded when/if expansion is enabled. 407 /// 408 /// If OwnsTokens is false, this method assumes that the specified stream of 409 /// tokens has a permanent owner somewhere, so they do not need to be copied. 410 /// If it is true, it assumes the array of tokens is allocated with new[] and 411 /// must be freed. 412 /// 413 void EnterTokenStream(const Token *Toks, unsigned NumToks, 414 bool DisableMacroExpansion, bool OwnsTokens); 415 416 /// RemoveTopOfLexerStack - Pop the current lexer/macro exp off the top of the 417 /// lexer stack. This should only be used in situations where the current 418 /// state of the top-of-stack lexer is known. 419 void RemoveTopOfLexerStack(); 420 421 /// EnableBacktrackAtThisPos - From the point that this method is called, and 422 /// until CommitBacktrackedTokens() or Backtrack() is called, the Preprocessor 423 /// keeps track of the lexed tokens so that a subsequent Backtrack() call will 424 /// make the Preprocessor re-lex the same tokens. 425 /// 426 /// Nested backtracks are allowed, meaning that EnableBacktrackAtThisPos can 427 /// be called multiple times and CommitBacktrackedTokens/Backtrack calls will 428 /// be combined with the EnableBacktrackAtThisPos calls in reverse order. 429 /// 430 /// NOTE: *DO NOT* forget to call either CommitBacktrackedTokens or Backtrack 431 /// at some point after EnableBacktrackAtThisPos. If you don't, caching of 432 /// tokens will continue indefinitely. 433 /// 434 void EnableBacktrackAtThisPos(); 435 436 /// CommitBacktrackedTokens - Disable the last EnableBacktrackAtThisPos call. 437 void CommitBacktrackedTokens(); 438 439 /// Backtrack - Make Preprocessor re-lex the tokens that were lexed since 440 /// EnableBacktrackAtThisPos() was previously called. 441 void Backtrack(); 442 443 /// isBacktrackEnabled - True if EnableBacktrackAtThisPos() was called and 444 /// caching of tokens is on. 445 bool isBacktrackEnabled() const { return !BacktrackPositions.empty(); } 446 447 /// Lex - To lex a token from the preprocessor, just pull a token from the 448 /// current lexer or macro object. 449 void Lex(Token &Result) { 450 if (CurLexer) 451 CurLexer->Lex(Result); 452 else if (CurPTHLexer) 453 CurPTHLexer->Lex(Result); 454 else if (CurTokenLexer) 455 CurTokenLexer->Lex(Result); 456 else 457 CachingLex(Result); 458 } 459 460 /// LexNonComment - Lex a token. If it's a comment, keep lexing until we get 461 /// something not a comment. This is useful in -E -C mode where comments 462 /// would foul up preprocessor directive handling. 463 void LexNonComment(Token &Result) { 464 do 465 Lex(Result); 466 while (Result.getKind() == tok::comment); 467 } 468 469 /// LexUnexpandedToken - This is just like Lex, but this disables macro 470 /// expansion of identifier tokens. 471 void LexUnexpandedToken(Token &Result) { 472 // Disable macro expansion. 473 bool OldVal = DisableMacroExpansion; 474 DisableMacroExpansion = true; 475 // Lex the token. 476 Lex(Result); 477 478 // Reenable it. 479 DisableMacroExpansion = OldVal; 480 } 481 482 /// LookAhead - This peeks ahead N tokens and returns that token without 483 /// consuming any tokens. LookAhead(0) returns the next token that would be 484 /// returned by Lex(), LookAhead(1) returns the token after it, etc. This 485 /// returns normal tokens after phase 5. As such, it is equivalent to using 486 /// 'Lex', not 'LexUnexpandedToken'. 487 const Token &LookAhead(unsigned N) { 488 if (CachedLexPos + N < CachedTokens.size()) 489 return CachedTokens[CachedLexPos+N]; 490 else 491 return PeekAhead(N+1); 492 } 493 494 /// RevertCachedTokens - When backtracking is enabled and tokens are cached, 495 /// this allows to revert a specific number of tokens. 496 /// Note that the number of tokens being reverted should be up to the last 497 /// backtrack position, not more. 498 void RevertCachedTokens(unsigned N) { 499 assert(isBacktrackEnabled() && 500 "Should only be called when tokens are cached for backtracking"); 501 assert(signed(CachedLexPos) - signed(N) >= signed(BacktrackPositions.back()) 502 && "Should revert tokens up to the last backtrack position, not more"); 503 assert(signed(CachedLexPos) - signed(N) >= 0 && 504 "Corrupted backtrack positions ?"); 505 CachedLexPos -= N; 506 } 507 508 /// EnterToken - Enters a token in the token stream to be lexed next. If 509 /// BackTrack() is called afterwards, the token will remain at the insertion 510 /// point. 511 void EnterToken(const Token &Tok) { 512 EnterCachingLexMode(); 513 CachedTokens.insert(CachedTokens.begin()+CachedLexPos, Tok); 514 } 515 516 /// AnnotateCachedTokens - We notify the Preprocessor that if it is caching 517 /// tokens (because backtrack is enabled) it should replace the most recent 518 /// cached tokens with the given annotation token. This function has no effect 519 /// if backtracking is not enabled. 520 /// 521 /// Note that the use of this function is just for optimization; so that the 522 /// cached tokens doesn't get re-parsed and re-resolved after a backtrack is 523 /// invoked. 524 void AnnotateCachedTokens(const Token &Tok) { 525 assert(Tok.isAnnotation() && "Expected annotation token"); 526 if (CachedLexPos != 0 && isBacktrackEnabled()) 527 AnnotatePreviousCachedTokens(Tok); 528 } 529 530 /// \brief Replace the last token with an annotation token. 531 /// 532 /// Like AnnotateCachedTokens(), this routine replaces an 533 /// already-parsed (and resolved) token with an annotation 534 /// token. However, this routine only replaces the last token with 535 /// the annotation token; it does not affect any other cached 536 /// tokens. This function has no effect if backtracking is not 537 /// enabled. 538 void ReplaceLastTokenWithAnnotation(const Token &Tok) { 539 assert(Tok.isAnnotation() && "Expected annotation token"); 540 if (CachedLexPos != 0 && isBacktrackEnabled()) 541 CachedTokens[CachedLexPos-1] = Tok; 542 } 543 544 /// \brief Specify the point at which code-completion will be performed. 545 /// 546 /// \param File the file in which code completion should occur. If 547 /// this file is included multiple times, code-completion will 548 /// perform completion the first time it is included. If NULL, this 549 /// function clears out the code-completion point. 550 /// 551 /// \param Line the line at which code completion should occur 552 /// (1-based). 553 /// 554 /// \param Column the column at which code completion should occur 555 /// (1-based). 556 /// 557 /// \returns true if an error occurred, false otherwise. 558 bool SetCodeCompletionPoint(const FileEntry *File, 559 unsigned Line, unsigned Column); 560 561 /// \brief Determine if this source location refers into the file 562 /// for which we are performing code completion. 563 bool isCodeCompletionFile(SourceLocation FileLoc) const; 564 565 /// \brief Instruct the preprocessor to skip part of the main 566 /// the main source file. 567 /// 568 /// \brief Bytes The number of bytes in the preamble to skip. 569 /// 570 /// \brief StartOfLine Whether skipping these bytes puts the lexer at the 571 /// start of a line. 572 void setSkipMainFilePreamble(unsigned Bytes, bool StartOfLine) { 573 SkipMainFilePreamble.first = Bytes; 574 SkipMainFilePreamble.second = StartOfLine; 575 } 576 577 /// Diag - Forwarding function for diagnostics. This emits a diagnostic at 578 /// the specified Token's location, translating the token's start 579 /// position in the current buffer into a SourcePosition object for rendering. 580 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID) { 581 return Diags->Report(FullSourceLoc(Loc, getSourceManager()), DiagID); 582 } 583 584 DiagnosticBuilder Diag(const Token &Tok, unsigned DiagID) { 585 return Diags->Report(FullSourceLoc(Tok.getLocation(), getSourceManager()), 586 DiagID); 587 } 588 589 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 590 /// token is the characters used to represent the token in the source file 591 /// after trigraph expansion and escaped-newline folding. In particular, this 592 /// wants to get the true, uncanonicalized, spelling of things like digraphs 593 /// UCNs, etc. 594 /// 595 /// \param Invalid If non-NULL, will be set \c true if an error occurs. 596 std::string getSpelling(const Token &Tok, bool *Invalid = 0) const; 597 598 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a 599 /// token is the characters used to represent the token in the source file 600 /// after trigraph expansion and escaped-newline folding. In particular, this 601 /// wants to get the true, uncanonicalized, spelling of things like digraphs 602 /// UCNs, etc. 603 static std::string getSpelling(const Token &Tok, 604 const SourceManager &SourceMgr, 605 const LangOptions &Features, 606 bool *Invalid = 0); 607 608 /// getSpelling - This method is used to get the spelling of a token into a 609 /// preallocated buffer, instead of as an std::string. The caller is required 610 /// to allocate enough space for the token, which is guaranteed to be at least 611 /// Tok.getLength() bytes long. The length of the actual result is returned. 612 /// 613 /// Note that this method may do two possible things: it may either fill in 614 /// the buffer specified with characters, or it may *change the input pointer* 615 /// to point to a constant buffer with the data already in it (avoiding a 616 /// copy). The caller is not allowed to modify the returned buffer pointer 617 /// if an internal buffer is returned. 618 unsigned getSpelling(const Token &Tok, const char *&Buffer, 619 bool *Invalid = 0) const; 620 621 /// getSpelling - This method is used to get the spelling of a token into a 622 /// SmallVector. Note that the returned StringRef may not point to the 623 /// supplied buffer if a copy can be avoided. 624 llvm::StringRef getSpelling(const Token &Tok, 625 llvm::SmallVectorImpl<char> &Buffer, 626 bool *Invalid = 0) const; 627 628 /// getSpellingOfSingleCharacterNumericConstant - Tok is a numeric constant 629 /// with length 1, return the character. 630 char getSpellingOfSingleCharacterNumericConstant(const Token &Tok, 631 bool *Invalid = 0) const { 632 assert(Tok.is(tok::numeric_constant) && 633 Tok.getLength() == 1 && "Called on unsupported token"); 634 assert(!Tok.needsCleaning() && "Token can't need cleaning with length 1"); 635 636 // If the token is carrying a literal data pointer, just use it. 637 if (const char *D = Tok.getLiteralData()) 638 return *D; 639 640 // Otherwise, fall back on getCharacterData, which is slower, but always 641 // works. 642 return *SourceMgr.getCharacterData(Tok.getLocation(), Invalid); 643 } 644 645 /// CreateString - Plop the specified string into a scratch buffer and set the 646 /// specified token's location and length to it. If specified, the source 647 /// location provides a location of the instantiation point of the token. 648 void CreateString(const char *Buf, unsigned Len, 649 Token &Tok, SourceLocation SourceLoc = SourceLocation()); 650 651 /// \brief Computes the source location just past the end of the 652 /// token at this source location. 653 /// 654 /// This routine can be used to produce a source location that 655 /// points just past the end of the token referenced by \p Loc, and 656 /// is generally used when a diagnostic needs to point just after a 657 /// token where it expected something different that it received. If 658 /// the returned source location would not be meaningful (e.g., if 659 /// it points into a macro), this routine returns an invalid 660 /// source location. 661 /// 662 /// \param Offset an offset from the end of the token, where the source 663 /// location should refer to. The default offset (0) produces a source 664 /// location pointing just past the end of the token; an offset of 1 produces 665 /// a source location pointing to the last character in the token, etc. 666 SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset = 0); 667 668 /// DumpToken - Print the token to stderr, used for debugging. 669 /// 670 void DumpToken(const Token &Tok, bool DumpFlags = false) const; 671 void DumpLocation(SourceLocation Loc) const; 672 void DumpMacro(const MacroInfo &MI) const; 673 674 /// AdvanceToTokenCharacter - Given a location that specifies the start of a 675 /// token, return a new location that specifies a character within the token. 676 SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,unsigned Char); 677 678 /// IncrementPasteCounter - Increment the counters for the number of token 679 /// paste operations performed. If fast was specified, this is a 'fast paste' 680 /// case we handled. 681 /// 682 void IncrementPasteCounter(bool isFast) { 683 if (isFast) 684 ++NumFastTokenPaste; 685 else 686 ++NumTokenPaste; 687 } 688 689 void PrintStats(); 690 691 /// HandleMicrosoftCommentPaste - When the macro expander pastes together a 692 /// comment (/##/) in microsoft mode, this method handles updating the current 693 /// state, returning the token on the next source line. 694 void HandleMicrosoftCommentPaste(Token &Tok); 695 696 //===--------------------------------------------------------------------===// 697 // Preprocessor callback methods. These are invoked by a lexer as various 698 // directives and events are found. 699 700 /// LookUpIdentifierInfo - Given a tok::identifier token, look up the 701 /// identifier information for the token and install it into the token. 702 IdentifierInfo *LookUpIdentifierInfo(Token &Identifier, 703 const char *BufPtr = 0) const; 704 705 /// HandleIdentifier - This callback is invoked when the lexer reads an 706 /// identifier and has filled in the tokens IdentifierInfo member. This 707 /// callback potentially macro expands it or turns it into a named token (like 708 /// 'for'). 709 void HandleIdentifier(Token &Identifier); 710 711 712 /// HandleEndOfFile - This callback is invoked when the lexer hits the end of 713 /// the current file. This either returns the EOF token and returns true, or 714 /// pops a level off the include stack and returns false, at which point the 715 /// client should call lex again. 716 bool HandleEndOfFile(Token &Result, bool isEndOfMacro = false); 717 718 /// HandleEndOfTokenLexer - This callback is invoked when the current 719 /// TokenLexer hits the end of its token stream. 720 bool HandleEndOfTokenLexer(Token &Result); 721 722 /// HandleDirective - This callback is invoked when the lexer sees a # token 723 /// at the start of a line. This consumes the directive, modifies the 724 /// lexer/preprocessor state, and advances the lexer(s) so that the next token 725 /// read is the correct one. 726 void HandleDirective(Token &Result); 727 728 /// CheckEndOfDirective - Ensure that the next token is a tok::eom token. If 729 /// not, emit a diagnostic and consume up until the eom. If EnableMacros is 730 /// true, then we consider macros that expand to zero tokens as being ok. 731 void CheckEndOfDirective(const char *Directive, bool EnableMacros = false); 732 733 /// DiscardUntilEndOfDirective - Read and discard all tokens remaining on the 734 /// current line until the tok::eom token is found. 735 void DiscardUntilEndOfDirective(); 736 737 /// SawDateOrTime - This returns true if the preprocessor has seen a use of 738 /// __DATE__ or __TIME__ in the file so far. 739 bool SawDateOrTime() const { 740 return DATELoc != SourceLocation() || TIMELoc != SourceLocation(); 741 } 742 unsigned getCounterValue() const { return CounterValue; } 743 void setCounterValue(unsigned V) { CounterValue = V; } 744 745 /// AllocateMacroInfo - Allocate a new MacroInfo object with the provide 746 /// SourceLocation. 747 MacroInfo* AllocateMacroInfo(SourceLocation L); 748 749 /// GetIncludeFilenameSpelling - Turn the specified lexer token into a fully 750 /// checked and spelled filename, e.g. as an operand of #include. This returns 751 /// true if the input filename was in <>'s or false if it were in ""'s. The 752 /// caller is expected to provide a buffer that is large enough to hold the 753 /// spelling of the filename, but is also expected to handle the case when 754 /// this method decides to use a different buffer. 755 bool GetIncludeFilenameSpelling(SourceLocation Loc,llvm::StringRef &Filename); 756 757 /// LookupFile - Given a "foo" or <foo> reference, look up the indicated file, 758 /// return null on failure. isAngled indicates whether the file reference is 759 /// for system #include's or not (i.e. using <> instead of ""). 760 const FileEntry *LookupFile(llvm::StringRef Filename, 761 bool isAngled, const DirectoryLookup *FromDir, 762 const DirectoryLookup *&CurDir); 763 764 /// GetCurLookup - The DirectoryLookup structure used to find the current 765 /// FileEntry, if CurLexer is non-null and if applicable. This allows us to 766 /// implement #include_next and find directory-specific properties. 767 const DirectoryLookup *GetCurDirLookup() { return CurDirLookup; } 768 769 /// isInPrimaryFile - Return true if we're in the top-level file, not in a 770 /// #include. 771 bool isInPrimaryFile() const; 772 773 /// ConcatenateIncludeName - Handle cases where the #include name is expanded 774 /// from a macro as multiple tokens, which need to be glued together. This 775 /// occurs for code like: 776 /// #define FOO <a/b.h> 777 /// #include FOO 778 /// because in this case, "<a/b.h>" is returned as 7 tokens, not one. 779 /// 780 /// This code concatenates and consumes tokens up to the '>' token. It 781 /// returns false if the > was found, otherwise it returns true if it finds 782 /// and consumes the EOM marker. 783 bool ConcatenateIncludeName(llvm::SmallString<128> &FilenameBuffer); 784 785private: 786 787 void PushIncludeMacroStack() { 788 IncludeMacroStack.push_back(IncludeStackInfo(CurLexer.take(), 789 CurPTHLexer.take(), 790 CurPPLexer, 791 CurTokenLexer.take(), 792 CurDirLookup)); 793 CurPPLexer = 0; 794 } 795 796 void PopIncludeMacroStack() { 797 CurLexer.reset(IncludeMacroStack.back().TheLexer); 798 CurPTHLexer.reset(IncludeMacroStack.back().ThePTHLexer); 799 CurPPLexer = IncludeMacroStack.back().ThePPLexer; 800 CurTokenLexer.reset(IncludeMacroStack.back().TheTokenLexer); 801 CurDirLookup = IncludeMacroStack.back().TheDirLookup; 802 IncludeMacroStack.pop_back(); 803 } 804 805 /// ReleaseMacroInfo - Release the specified MacroInfo. This memory will 806 /// be reused for allocating new MacroInfo objects. 807 void ReleaseMacroInfo(MacroInfo* MI); 808 809 /// ReadMacroName - Lex and validate a macro name, which occurs after a 810 /// #define or #undef. This emits a diagnostic, sets the token kind to eom, 811 /// and discards the rest of the macro line if the macro name is invalid. 812 void ReadMacroName(Token &MacroNameTok, char isDefineUndef = 0); 813 814 /// ReadMacroDefinitionArgList - The ( starting an argument list of a macro 815 /// definition has just been read. Lex the rest of the arguments and the 816 /// closing ), updating MI with what we learn. Return true if an error occurs 817 /// parsing the arg list. 818 bool ReadMacroDefinitionArgList(MacroInfo *MI); 819 820 /// SkipExcludedConditionalBlock - We just read a #if or related directive and 821 /// decided that the subsequent tokens are in the #if'd out portion of the 822 /// file. Lex the rest of the file, until we see an #endif. If 823 /// FoundNonSkipPortion is true, then we have already emitted code for part of 824 /// this #if directive, so #else/#elif blocks should never be entered. If 825 /// FoundElse is false, then #else directives are ok, if not, then we have 826 /// already seen one so a #else directive is a duplicate. When this returns, 827 /// the caller can lex the first valid token. 828 void SkipExcludedConditionalBlock(SourceLocation IfTokenLoc, 829 bool FoundNonSkipPortion, bool FoundElse); 830 831 /// PTHSkipExcludedConditionalBlock - A fast PTH version of 832 /// SkipExcludedConditionalBlock. 833 void PTHSkipExcludedConditionalBlock(); 834 835 /// EvaluateDirectiveExpression - Evaluate an integer constant expression that 836 /// may occur after a #if or #elif directive and return it as a bool. If the 837 /// expression is equivalent to "!defined(X)" return X in IfNDefMacro. 838 bool EvaluateDirectiveExpression(IdentifierInfo *&IfNDefMacro); 839 840 /// RegisterBuiltinPragmas - Install the standard preprocessor pragmas: 841 /// #pragma GCC poison/system_header/dependency and #pragma once. 842 void RegisterBuiltinPragmas(); 843 844 /// RegisterBuiltinMacros - Register builtin macros, such as __LINE__ with the 845 /// identifier table. 846 void RegisterBuiltinMacros(); 847 848 /// HandleMacroExpandedIdentifier - If an identifier token is read that is to 849 /// be expanded as a macro, handle it and return the next token as 'Tok'. If 850 /// the macro should not be expanded return true, otherwise return false. 851 bool HandleMacroExpandedIdentifier(Token &Tok, MacroInfo *MI); 852 853 /// isNextPPTokenLParen - Determine whether the next preprocessor token to be 854 /// lexed is a '('. If so, consume the token and return true, if not, this 855 /// method should have no observable side-effect on the lexed tokens. 856 bool isNextPPTokenLParen(); 857 858 /// ReadFunctionLikeMacroArgs - After reading "MACRO(", this method is 859 /// invoked to read all of the formal arguments specified for the macro 860 /// invocation. This returns null on error. 861 MacroArgs *ReadFunctionLikeMacroArgs(Token &MacroName, MacroInfo *MI, 862 SourceLocation &InstantiationEnd); 863 864 /// ExpandBuiltinMacro - If an identifier token is read that is to be expanded 865 /// as a builtin macro, handle it and return the next token as 'Tok'. 866 void ExpandBuiltinMacro(Token &Tok); 867 868 /// Handle_Pragma - Read a _Pragma directive, slice it up, process it, then 869 /// return the first token after the directive. The _Pragma token has just 870 /// been read into 'Tok'. 871 void Handle_Pragma(Token &Tok); 872 873 /// EnterSourceFileWithLexer - Add a lexer to the top of the include stack and 874 /// start lexing tokens from it instead of the current buffer. 875 void EnterSourceFileWithLexer(Lexer *TheLexer, const DirectoryLookup *Dir); 876 877 /// EnterSourceFileWithPTH - Add a lexer to the top of the include stack and 878 /// start getting tokens from it using the PTH cache. 879 void EnterSourceFileWithPTH(PTHLexer *PL, const DirectoryLookup *Dir); 880 881 /// IsFileLexer - Returns true if we are lexing from a file and not a 882 /// pragma or a macro. 883 static bool IsFileLexer(const Lexer* L, const PreprocessorLexer* P) { 884 return L ? !L->isPragmaLexer() : P != 0; 885 } 886 887 static bool IsFileLexer(const IncludeStackInfo& I) { 888 return IsFileLexer(I.TheLexer, I.ThePPLexer); 889 } 890 891 bool IsFileLexer() const { 892 return IsFileLexer(CurLexer.get(), CurPPLexer); 893 } 894 895 //===--------------------------------------------------------------------===// 896 // Caching stuff. 897 void CachingLex(Token &Result); 898 bool InCachingLexMode() const { 899 // If the Lexer pointers are 0 and IncludeMacroStack is empty, it means 900 // that we are past EOF, not that we are in CachingLex mode. 901 return CurPPLexer == 0 && CurTokenLexer == 0 && !IncludeMacroStack.empty(); 902 } 903 void EnterCachingLexMode(); 904 void ExitCachingLexMode() { 905 if (InCachingLexMode()) 906 RemoveTopOfLexerStack(); 907 } 908 const Token &PeekAhead(unsigned N); 909 void AnnotatePreviousCachedTokens(const Token &Tok); 910 911 //===--------------------------------------------------------------------===// 912 /// Handle*Directive - implement the various preprocessor directives. These 913 /// should side-effect the current preprocessor object so that the next call 914 /// to Lex() will return the appropriate token next. 915 void HandleLineDirective(Token &Tok); 916 void HandleDigitDirective(Token &Tok); 917 void HandleUserDiagnosticDirective(Token &Tok, bool isWarning); 918 void HandleIdentSCCSDirective(Token &Tok); 919 920 // File inclusion. 921 void HandleIncludeDirective(Token &Tok, 922 const DirectoryLookup *LookupFrom = 0, 923 bool isImport = false); 924 void HandleIncludeNextDirective(Token &Tok); 925 void HandleIncludeMacrosDirective(Token &Tok); 926 void HandleImportDirective(Token &Tok); 927 928 // Macro handling. 929 void HandleDefineDirective(Token &Tok); 930 void HandleUndefDirective(Token &Tok); 931 932 // Conditional Inclusion. 933 void HandleIfdefDirective(Token &Tok, bool isIfndef, 934 bool ReadAnyTokensBeforeDirective); 935 void HandleIfDirective(Token &Tok, bool ReadAnyTokensBeforeDirective); 936 void HandleEndifDirective(Token &Tok); 937 void HandleElseDirective(Token &Tok); 938 void HandleElifDirective(Token &Tok); 939 940 // Pragmas. 941 void HandlePragmaDirective(); 942public: 943 void HandlePragmaOnce(Token &OnceTok); 944 void HandlePragmaMark(); 945 void HandlePragmaPoison(Token &PoisonTok); 946 void HandlePragmaSystemHeader(Token &SysHeaderTok); 947 void HandlePragmaDependency(Token &DependencyTok); 948 void HandlePragmaComment(Token &CommentTok); 949 void HandlePragmaMessage(Token &MessageTok); 950 // Return true and store the first token only if any CommentHandler 951 // has inserted some tokens and getCommentRetentionState() is false. 952 bool HandleComment(Token &Token, SourceRange Comment); 953}; 954 955/// \brief Abstract base class that describes a handler that will receive 956/// source ranges for each of the comments encountered in the source file. 957class CommentHandler { 958public: 959 virtual ~CommentHandler(); 960 961 // The handler shall return true if it has pushed any tokens 962 // to be read using e.g. EnterToken or EnterTokenStream. 963 virtual bool HandleComment(Preprocessor &PP, SourceRange Comment) = 0; 964}; 965 966} // end namespace clang 967 968#endif 969